|
|
System Analysis Advisory Committee Meeting Notes
March 31, 2003 - 9:30 a.m. - 12:00 p.m.
NORTHWEST POWER PLANNING COUNCIL OFFICES
PORTLAND, OREGON
DRAFT
I. Greetings, Introductions and Review of the Agenda.
The
March 31, 2003 System Analysis Advisory Committee meeting, held at the
Northwest Power Planning Council's offices in Portland, Oregon, was
chaired by Michael Schilmoeller of the Council staff.
The
following is a distillation (not a verbatim transcript) of items discussed
during the call, together with actions taken on those items. Please
note that some enclosures referenced in the body of the text may be too
lengthy to attach; all enclosures referenced are available upon request
from Schilmoeller at 503/820-2314.
Schilmoeller welcomed everyone to today's meeting, led a round of
introductions, then reviewed today's agenda. Schilmoeller noted that
copies of his presentation are available via the NWPPC website; please
refer to this document for full details, including graphs and
charts.
The
notes from the SAAC's February 27 meeting were approved as written.
2. The Milestones for Assessing Risk.
Schilmoeller went through some of the Power Plan issues that are
influenced by risk assessment, as well as the sources of that risk ?
fuel price and availability, load uncertainty, hydrogeneration
uncertainty, transmission congestion and a number of other factors. He
touched on the temporal aspect of risk (short-term variation vs. long-term
uncertainty) and some potential ways to mitigate those risks, including
optionality vs hedging.
Schilmoeller went on to enumerate some risk assessment issues, potential
ways to address risk, how risk might be determined for each portfolio, and
the preferred objective mathematical function. He described the Contingent
Value At Risk (CVAR) function, which evaluates both cost and risk, then
moved on to some of the questions this portion of the portfolio analysis
is intended to answer, and how he would propose to answer them (by
building the regional assessment from the bottom up).
Next, Schilmoeller went through a flow chart diagram illustrating his
proposed risk evaluation method. He then provided an example, starting
with wind, to illustrate this method. After a few minutes of discussion,
Schilmoeller moved on to Olivia, which he described as a ?meta-model?
that will allow individual users to create their own models to evaluate
the risks to their particular system. He spent a few minutes demonstrating
how the Olivia model works.
We
talked about futures this morning, and how we're going to pick variables
for the next 20 years, said John Fazio ? how are those introduced to
Olivia? In Olivia herself, basically anything that is a stochastic
variable, such as loads, points to a behavior set, Schilmoeller replied.
It is a monthly model; within that month you will see hourly variation,
and that's where futures would be reflected. So when you execute Olivia,
you would run, say, 500 games, and Olivia will choose the random
variables, then calculate the cost? Fazio asked. Not quite, Schilmoeller
replied ? each game is a 20-year period. Olivia tells the workbook how
those variables relate to one another, added Marty Howard.
3. Futures.
Schilmoeller began this agenda item with a review of Council load
forecasts over the past 20 years, using a graph titled ?Historical
Forecasts Compared to Actual Firm Sales.? He noted that, by and large,
the Council has done a reasonably good job of predicting actual loads.
Schilmoeller also touched on the Council's historic oil price forecasts,
noting that the goal of this portion of the presentation was to provide
some historical perspective to the discussion of futures in the context of
the portfolio model.
Schilmoeller then moved on to the Council's current forecasts of
regional loads, natural gas prices, hydro generation, carbon tax and
wholesale power prices out to 2021. He noted that one of the primary
purposes of this portion of the portfolio analysis is to demonstrate the
interrelationships between these factors; to that end, he has developed an
animation, using various potential futures and random price and load
spikes, to illustrate those relationships via a series of changing graphs.
He added that his purpose today was not to defend this phase of the
analysis, but to take any comments the other SAAC participants might have.
The
group offered a variety of clarifying questions and comments; one
participant noted that one major variable that is not explicitly addressed
in this analysis is what's happening in the rest of the west, both in
terms of prices and loads. Schilmoeller paused his ?movie? at one
potential future; Jim Litchfield noted that, in his opinion, this
particular future, which includes medium loads, low natural gas prices,
generally good water years but high wholesale power prices, is extremely
unlikely. The group discussed some potential factors that could cause such
a future, including underbuilding of resources -- in particular, outside
the region -- or a long-term credit crisis. Another participant noted that
the two spikes on the wholesale power price graph correspond with low
water years, but what seems unrealistic, to him, is the fact that power
prices stay high for many years in between those two ?horns.?
Basically, what we're hearing is that this particular future may be
extremely unlikely, but it could happen, Kevin Nordt observed.
Overall, the sense in the room seems to be that they would like to see a
bit more correlation to the basic drivers, said Terry Morlan ? when gas
prices are high, then power prices should intuitively be high, for
example. Basically, if it's a one in a million chance, it becomes almost
irrelevant, another participant observed.
The
presentation paused at another scenario in which there were a number of
poor water years, coupled with steadily-rising natural gas prices and
medium loads, yet wholesale power prices stayed low through the entire
period of the analysis. Again, that is counter-intuitive, Litchfield
observed; there must be something driving your model that is not
inherently obvious in these results. If what you?re looking for is
feedback about how these variables interact, he said, to me, your model
outputs don't seem to match very well with what I, at least, would
expect to see.
Litchfield added that, in his opinion, this analysis is going to be a hard
sell to the Council; given its complexity, it may or may not increase
their comfort level in assessing future risk, and correspondingly may or
may not be useful to the Council in developing a Power Plan that
effectively addresses future load growth and resource balance in the
region. that's useful feedback, Schilmoeller said. Perhaps by
making the market price more directly tied to the other parameters, the
results will become more intuitive.
4. Analysis of Data.
Howard went through a series of points relating to gas and electricity
prices, including portfolio model inputs (Sumas gas, Mid-Columbia
electric). He touched on what is available on these two factors, in terms
of the historical database, then moved on to demonstrate the covariance
matrix he has created to plot that data. The group offered a few minor
clarifying questions and comments. Howard said his next step will be to
generate a series of electric prices that demonstrate the same
characteristics -- behave in the same way -- as the real historic data.
Among his conclusions was the fact that volatility in Mid-Columbia
electric prices appears to be varying, in a predictable pattern. The next
thing to do is to capture, with a relatively simple regression, the
influence of other variables, such as hydro operations, Howard said.
So
essentially what you've done here is to subtract out the influence of
gas prices on Mid-Columbia electricity prices, and this is what is left?
Schilmoeller asked. that's correct, Howard replied.
5. Next SAAC Meeting Date.
The
next meeting of the System Analysis Advisory Committee was set for April
29. Meeting summary prepared by Jeff Kuechle, NWPPC contractor.
|
home